Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

1745

July 11th, 2016 14:00

Add an IP to MDM/TB servers?

Hello,

I have a mixed system with ESX and Windows servers in it. It was built with the ESX installer and all is working well for the most part, however all log collections fail to the Windows hosts that are not on the same network as what is associated with the MDMs and TB SVMs. The MDMs and TB see a 10.1.3.x network in cluster query and have a mgmt IP of 10.5.1.x.

The hosts that fail log collection through the GW URL are not on the 10.1.3.x network so after some troubleshooting, I have come to the conclusion that the reason for the failure is because they are trying to talk to the MDM/TB IP's, which they can't communicate with. Is there a way to add another IP (in this case my mgmt IP's of the SVMs) to the MDM/TB IP?

Looking at the docs, all the --query_cluster outputs show the MDM/TB's with both IP's, so that is what I am trying to attain.

thanks!

306 Posts

July 11th, 2016 22:00

Hi Keith,

Check out:

https://support.emc.com/kb/484205

Thanks,

Pawel

522 Posts

July 12th, 2016 05:00

Thanks Pawel.....so much for my support.emc.com searching skills! not sure how I missed that one. I will try this out to change the MDM IP's and let you know how I make out...thanks for the response!

522 Posts

July 12th, 2016 12:00

haha...I just fixed that...sorry! You definitely deserve the credit.

The two that fail are an ESX SDC (an ESX host that is running the TB SVM) and another Windows host on another network, but that has IP connectivity to the IP's associated with the MDM/TB SVMs,

The error is vague - just says it timed out - doesn't give any other reason in the IM GUI.

Thanks!

306 Posts

July 12th, 2016 12:00

Hi Keith,

Just for clarification, are those ESXi SDCs or some other OS'es? What exactly is the error messages you are getting in the IM ("Details").

BTW you marked wrong answer as "correct", haha

Pawel

522 Posts

July 12th, 2016 12:00

Thanks Pawel...this worked like a charm...not sure how my search missed it. I changed this and my log collection still fails for 2 hosts that seem to be trying to communicate to all IP's out there and since it can't get to the SDS-only network, it fails...which is odd. I'll have to continue to troubleshoot this and figure out what the log collection is trying to hit when using the GW to collect the logs since the IP's of the SDC's have network communication to the MDMs/TB so not sure what is up there. Thank you again for the help above!!

-Keith

522 Posts

July 12th, 2016 13:00

Thanks Pawel.....so that KB article I did see..haha. and i have ignored the SDC on the ESX host for the most part. What has been consuming me is the random Windows host I added to the cluster as an SDS and SDC (to make it a mixed cluster). I installed the LIA and approved the certificate, changed passwords, etc so that communication is all set and that particular host still seems to fail. When I do a netstat on it, it looks as if it is establishing connections with every IP in the ScaleIO config and since the networks of this Windows VM are different than the ESX hosts (routable and pingable, but not on the same subnet for example), I am expecting them to be able to generate the logs - is that a false expectation from my side? I would imagine that IP communication (assuming all components are installed) to the MDM IP's would suffice to call the script to collect the logs....is that a correct assumption? Just not sure if I am missing something about the communication needed for log collection. If I can get an SDS added over a different network, I was assuming logs would be collected

Also, you are correct, I have been using the GW through the plugin, but also just pulling it up in a web browser to see if there was any difference.

Thanks again...you have been very informative in your answers and I appreciate it.

306 Posts

July 12th, 2016 13:00

Hi Keith,

Haha, it's not about the credit, rather making the correct answer easier to to find by someone else

If they are ESX SDCs, then this behavior is normal - on the ESXi we can't really run "get_info" script (in fact, there isn't even one installed) - please see another KB here:

https://support.emc.com/kb/485247

In general, you should be opening ScaleIO Gateway from the SIO Web Plugin - then it will "mask" all the ESXi SDCs from the query and won't generate any errors. If you open the GW directly, it is not aware of the fact these are ESXi and will try to open a connection to them to run get_info - and eventually fail, as you can't do that in ESX.

Hope that helps! Let me know if you have any other questions.

Pawel

306 Posts

July 13th, 2016 00:00

Hi Keith,

No problem, I am happy I could assist

I'm wondering if there might be firewall enabled on that Windows host, blocking access to the LIA agent? And/or a firewall in between these two networks that would prevent GW->LIA traffic?

In general, you should be able to collect get_info from a Windows box, something must be on its way.. can you try to telnet from the GW box to tcp/9099 on the Windows box?

Thanks!

Pawel

522 Posts

July 13th, 2016 10:00

Hi Pawel, I am finally running clean on the log collections and setup now that I changed the IP's and fixed the LIA's on the hosts. I had to uninstall and re-install the LIA and then set the password and it works. I'm using the nonsecure method right now. The error messages in the logs are tough to decipher,but I got a crash course in them this week!

I did have one question related to the stuff above - is it normal to have the GW GUI launched with the ScaleIO plugin show the "Install" pane, but have a GUI launched to the same IP outside of the plugin NOT show that pane? I noticed that now that I am getting ready to test some additions/removals.

I think I am all set with this thread and if I have any more issues moving forward, I will post another thread...thank you very much for your help.

-Keith

No Events found!

Top